Get Started
Govern an agent end to end in a sandbox.Govern your first agent
Allow employee lookups, redact contact details, and deny payroll — from Cursor, Claude, or curl.
Call a model through a pool
Route chat through a model pool with failover, authorization, and your provider key held server-side.
Build a governed app
Wire model and tool calls in one app, then click through allow, redact, and deny.
Govern per end user
The agent inherits each user’s permissions — same agent, different tool access and answers per person.
Role
Learn what developer, security, operator, and platform admin each own on Dome — guided sandbox walks for each persona.Developer
Register an agent, call tools through a Gateway, and confirm allow and deny in audit.
Security
Author and simulate Cedar, redact sensitive fields, prove denials, and export evidence.
Operator
Stand up the substrate: attach tools, expose a Gateway, grant access, and watch audit.
Platform Admin
Set who may administer Dome, check platform health, and export an audit snapshot.
Production
Diagnose a live denial, put an existing LLM client under Dome, deploy a working agent, or use OIDC or bound act-as.Debug a tool call denial
Gateway admission, Cedar, and audit.
Adopt an existing app
Repoint an LLM client at Dome without rewriting logic.
Deploy your agent
Docker, Render, Fly, and Lambda — coming soon.
Use OIDC or bound act-as
Replace sandbox HMAC act-as with an IdP JWT or a bound enrollment identity — coming soon.
Examples
Working examples you can adapt:Interactive agent
Person in the loop — coming soon.
Noninteractive agent
Batch or background — coming soon.
Use cases
Code Execution Agent
Governed code execution.
Multi-Adapter Tools Agent
Multiple tool adapters behind one agent.
Knowledge Base Agent
Retrieval over a knowledge base.